Predicted outcomes for patients with luminal valve implantation

Abstract

Various aspects of the methods, systems, and use cases for training a model to determine whether a patient is a candidate for fitting an intraluminal valve based on collateral ventilation data may be used. The method may include receiving sensor data based on pressure or airflow at a target portion of the patient's lung that is obstructed such that it cannot receive air through the respiratory airway of the lung. The method may include training a machine learning model based at least in part on the training data (e.g., based on the sensor data) to predict a patient's respiratory outcome via an indication of whether collateral ventilation is present at the target lung portion of the particular patient.

Description

Technical Field

[0001] Claim of Priority This application claims the benefit of priority to U.S. Provisional Application No. 63 / 262776, entitled "ENDOLUMINAL VALVE PLACEMENT PATIENT OUTCOME PREDICTION," filed on October 20, 2021, which is hereby incorporated by reference in its entirety.

Background Art

[0002] An endoluminal valve can be placed inside an airway leading to an affected part of the lung to redirect breathed air from the affected area to a healthier part of the lung. These endoluminal valves are check valves that prevent breathed air from entering the affected part of the lung while allowing air and bodily fluids (e.g., mucus) to escape from the affected part of the lung. The affected part of the lung (e.g., a lung region with severe emphysema) tends to increase in volume, preventing other parts from expanding properly. Thus, endoluminal valve placement is an effective treatment for reducing the volume occupied by the affected part of the lung (which does not contribute to O2-CO2 gas exchange). Reducing the volume of the affected part provides more space for the healthy parts of the lung to fully expand during the breathing cycle, thereby enabling significantly more gas exchange. Unfortunately, some affected parts of the lung receive good airflow from collateral ventilation, in which gas passes from one lung unit to an adjacent lung unit through collateral channels such as alveolar pores and / or direct airway anastomoses. Endoluminal valve placement can be an effective treatment even when there is some degree of collateral ventilation, but when the degree of collateral ventilation is relatively high, endoluminal valve placement can be an ineffective treatment for reducing lung volume.

Summary of the Invention

Means for Solving the Problems

[0003] While the drawings are not necessarily drawn to scale, similar numbers in different drawings may represent similar components. Similar numbers with different letter suffixes may represent different instances of similar components. The drawings generally illustrate the various embodiments discussed herein as examples, not limitations. [Brief explanation of the drawing]

[0004] [Figure 1] This figure shows an exemplary collateral ventilation quantification system (CVQS) according to at least one example of the present disclosure. [Figure 2] This is a training diagram of a machine learning model, based on at least one example of the present disclosure. [Figure 3] This is a machine learning model inference diagram based on at least one example of the present disclosure. [Figure 4A] This is a flow and pressure diagram illustrating collateral ventilation, according to at least one example of the present disclosure. [Figure 4B] This is a flow and volume diagram illustrating collateral ventilation according to at least one example of the present disclosure. [Figure 4C] This is a flow and pressure diagram illustrating the absence of collateral ventilation (CV) in at least one example of the present disclosure. [Figure 4D] This is a flow and volume diagram showing the absence of collateral ventilation (CV) in at least one example of the present disclosure. [Figure 5] A flowchart illustrating a technique for training a model to determine, based on collateral ventilation data, whether a patient is a candidate for lumen valve implantation, as illustrated by at least one example of the present disclosure. [Figure 6] This is a block diagram of an exemplary machine in which one or more of the techniques discussed herein may be performed, according to at least one example of this disclosure. [Modes for carrying out the invention]

[0005] As mentioned above, some affected areas of the lung may receive airflow through collateral ventilation and / or collateral channels. Depending on the extent of this, this collateral ventilation (CV) may render lumen valve placement an ineffective treatment for lung volume loss. This is because while the lumen valve may function properly and prevent air from entering the affected area of ​​the lung through the normal airway (e.g., bronchi), collateral ventilation may be present to the extent that air can freely enter the affected area without passing through the normal airway in which the valve is placed. Therefore, before treating patients with chronic obstructive pulmonary disease (COPD) with lumen valve placement, the target lung region is usually evaluated to ensure that the COPD patient is not receiving airflow through collateral ventilation to a degree that makes it likely that the patient will not respond well to lumen valve placement.

[0006] A collateral ventilation quantification system (CVQS) may be used to assess the target lung region for central venous (CV). The CVQS may output sensor data (e.g., pressure and / or airflow) on a graphical user interface (GUI), etc. In certain non-limiting examples, measurements of pressure within an obstructed lobe and airflow entering (or leaving) the obstructed lobe may be sampled periodically or continuously. The output may include graphs and / or data points over time. A clinician (e.g., physician, surgeon, specialist, etc.) may evaluate the sensor data to determine whether CV is present in the target lung region and / or the extent of CV. However, in some cases, the clinician may not be able to determine whether CV is present, it may take a long time to assess whether CV is present, and / or they may not be able to accurately determine the extent of any CV present with high reliability. In some cases, even if CV is present to some extent, the patient may still benefit from lumen valve implantation, depending on various factors such as the patient's age, weight, body mass index (BMI), and medical history, to name a few. However, the likelihood of a favorable outcome from luminal valve implantation cannot be determined solely by the clinician's assessment.

[0007] The systems and techniques described herein provide models (e.g., classifiers, trained models, etc., such as using machine learning techniques also known as artificial intelligence) to provide information relating to whether a patient has a central venous cavity (CV). The models may output an indication that a CV is present (CV+) or absent (CV-) in a target lung region of the patient. In some examples, the models may output the probability and / or confidence level of whether a CV is present, the degree of the CV (e.g., estimates of CV flow rates such as low, medium, and high, flow rate results over time, etc.), the degree of collateral resistance between the obstructed lung portion and adjacent lung portions, etc.

[0008] In one example, the model might output a mark indicating whether the patient is a suitable candidate for a lumen valve (in addition to, or instead of, whether a CV is present). In this example, the mark might include a CV determination, but it doesn't have to depend entirely on whether a CV is present. For example, a patient might benefit from a lumen valve even if, in some cases, they have several CVs.

[0009] Figure 1 shows an exemplary collateral ventilation quantification system (CVQS) 100 according to at least one example of the present disclosure. While the CVQS 100 is shown as a positive pressure system, other examples may not include positive pressure. For example, the illustrated CVQS 100 includes airflow into a lung portion, while other examples may include a system that allows airflow from a lung portion while preventing airflow into the lung portion. Generally speaking, in some examples, airflow into and out of a lung portion is restricted to determine whether collateral ventilation is present in that portion.

[0010] CVQS100 includes a CVQS device 102 and a CVQS tubing kit 104. CVQS device 102 includes a flow meter 106, a pressure gauge 108, a display device 110 (e.g., including a graphical user interface), and a constant-pressure air supply device 112 (e.g., continuous positive airway pressure (CPAP)). In some examples, CVQS device 102 may omit one or more of these components. For example, the constant-pressure air supply device 112 may not be used (and optionally, the flow meter 106 and pressure gauge 108 may be omitted in this example). In one example, the display device 110 may be located remotely from CVQS device 102 (e.g., connected communicably via a wired or wireless communication architecture such as Ethernet, Wi-Fi, or Bluetooth).

[0011] The CVQS tubing kit 104 may include various tubes and / or filters, such as tubes used to add air to the lung portion and / or tubes to remove air from the lung portion. The CVQS tubing kit 104 may include a check valve 114 that can restrict the inflow and outflow of air to the lung portion. In the example shown in CVQS 100, the check valve 114 blocks the airflow from the lung portion but allows air to flow into the lung portion (e.g., via a constant-pressure air supply device 112). In other examples, the check valve may prevent airflow to the lung portion and allow air to exit the lung portion.

[0012] The CVQS 100 shown in Figure 1 supplies positive pressure to the obstructed portion of the lung and measures the pressure over time (using a pressure gauge 108) and / or the flow rate over time (using a flow meter 106). The pressure over time and / or flow rate may be used to assess whether the obstructed portion of the lung is ventilating through collateral vessels. The target lung portion may be obstructed using an obstruction balloon (e.g., Olympus balloon catheter B7-2C), a lumen valve, and / or other suitable obstruction device 116. In one example, a balloon catheter temporarily isolates part or more portions of the lung by inflating in the airway. While inflating in the airway, the CVQS device 102 may provide an airflow through the lumen of the catheter, for example, at a constant pressure (e.g., 10 cmH2O). The airflow through the catheter lumen may be monitored by the flow meter 106 and / or pressure gauge 108 of the CVQS device 102, and data (e.g., flow rate, pressure, and / or total volume) may be output. The output may be displayed on a display device 110 (e.g., a portable computer, mobile device, etc.). In some examples, the output is stored without being displayed. The output may be used to predict the patient's condition (e.g., whether collateral ventilation is present and / or whether the lumen valve is a suitable or unsuitable candidate) via a machine learning-trained model, etc.

[0013] The tubing kit 104 may connect the balloon catheter to the CVQS device 102. The tubing kit 104 may include a filter and a check valve to allow only airflow to the target lobe (e.g., supplied by the constant-pressure air supply device 112) for a period of time such as 1 to 10 minutes (e.g., 3 to 5 minutes) or 5 to 20 minutes (e.g., about 10 minutes). The flow through the balloon catheter may be affected by the amount of collateral ventilation in the tissue. For example, air may continue to flow into the obstructed lobe until the end-expiratory pressure equals the CVQS source pressure. If collateral ventilation is present, air may continue to flow into the lobe as it escapes through the collateral pathway (e.g., it may not reach the end pressure, or the flow may continue even if it is reached). If air can escape from the obstructed lobe to an adjacent lobe through the collateral ventilation channel, the pressure in the lobe may not be uniform, and / or air may continue to flow into the obstructed lobe through the lumen of the obstruction catheter, etc., and / or out of the obstructed lobe through the collateral ventilation channel.

[0014] In some patients with emphysema or other lung conditions, the alveoli of the lungs may swell, preventing gas exchange. As the alveoli enlarge, they compress other lobes, reducing the amount of air exchanged. A luminal valve may be used to allow air to escape from a portion of the lung but not enter, thereby reducing the lung volume of the affected portion. This type of valve can alleviate some of the problems associated with certain symptoms in these patients. However, if collateral ventilation is present, air leakage between lobes through collateral channels can cause the valve to malfunction or its effectiveness to decrease. Collateral ventilation can occur when air moves between lobes and / or between segments of a lobe. Collateral ventilation can occur in patients with emphysema. In some patients, a crack in the lung ruptures, creating a breakthrough between lobes, resulting in collateral ventilation. In some cases, a crack integrity score may be used to assess whether a patient is a suitable candidate for a luminal valve. For example, if a patient's crack integrity score exceeds approximately 90%, the lobe may be considered sufficiently intact to allow for valve placement within the patient's body. If the score is below 90%, there may be greater uncertainty about whether the luminal valve is effective. The CVQS100 can be used in conjunction with machine learning-trained models to assess patients and determine whether they would benefit from a luminal valve.

[0015] Figure 2 is a training diagram of a machine learning model 202, according to at least one example of the present disclosure. Figure 200 shows the components and inputs for training model 202 using machine learning.

[0016] Machine learning (ML) is an application that provides computer systems with the ability to perform tasks without explicit programming by making inferences based on patterns found in data analysis. Machine learning explores the study and construction of algorithms, also referred to herein as tools, that can learn from existing data and make predictions about new data. While examples of several machine learning tools are presented, the principles described herein can also be applied to other machine learning tools.

[0017] Machine learning algorithms use data (such as action primitives and / or interaction primitives, target vectors, rewards, etc.) to find the correlations between identified features that affect the results. A feature is an individually measurable property of the observed phenomenon. Exemplary features of model 202 can include diagnostic data (e.g., from a physician), reported patient outcome data, and / or other labels related to the patient's condition and / or situation, regardless of the presence or absence of an intracavitary valve. Features that can include label data and / or can be called label data can be compared with input data such as pressure data, flow rate data, etc.

[0018] The concept of features is related to the concept of explanatory variables used in statistical techniques such as linear regression. Selecting beneficial, discriminatory, and independent features is important for the effective operation of ML in pattern recognition, classification, and regression. Features can be of various types, such as numerical features, strings, and graphs.

[0019] During training, the ML algorithm analyzes the input data based on the identified features and optionally defined configuration parameters for training (such as patient data like environmental data, state data, demographics, and / or co-morbidities). The result of training is model 202, which can receive inputs to generate complex tasks.

[0020] For example, input data may be labeled (e.g., for use as features in a training phase). Labeling may include identifying the patient's state and / or condition after and / or without intervention. For example, the patient's state and / or condition may be labeled as including or not including a lumen valve intervention. The patient's state and / or condition may include objective outcomes (e.g., whether or not a CV was present, whether the patient's respiration improved based on objective tests) and / or subjective outcomes (e.g., the patient perceives an improvement in respiration and / or quality of life, and the clinician assesses whether a CV was present, for example, by using visual determination). Outcomes may be identified over time, such as 3 months and / or 6 months after (or without) intervention. Time labels may be weighted and / or used to generate different versions of Model 202. An example of an objective outcome may include a crack integrity score. Scores may be weighted, such as ≥90 indicating no CV and ≤80 indicating CV. In some examples, labels may include weighting for the degree of CV. In these examples, weightings may be based on low or high flow rates to improve Model 202. Some of the results described herein may be used to update Model 202 after initial training.

[0021] The input training data for Model 202 may include pressure data and / or flow rate data, as discussed above. Other data used for input may include CT scans (e.g., high resolution) with corresponding crack integrity scores, disease status, patient age, infection, and other confounding factors.

[0022] A neural network, also called an artificial neural network, is a computing system based on the consideration of the biological neural network of an animal's brain. Such a system gradually improves its performance for performing a task called learning, usually without performing task-specific programming. For example, in image recognition, a neural network is trained to identify an image containing an object by analyzing exemplary images in which the names of the objects are tagged. Once the objects and names are learned, the analysis results can be used to identify and / or classify the objects in an untagged image. For example, in FIG. 2, the model 202 can be trained to identify whether a patient has a CV based on input data (e.g., pressure and / or flow rate), and / or classify the patient (e.g., as a candidate for an endovascular valve or not, and / or along with the probability and / or confidence of success of an endovascular valve procedure).

[0023] A neural network is based on a collection of connected units called neurons, and each connection, called a synapse between neurons, can transmit a one-way signal with different activation strengths depending on the strength of the connection. The receiving neuron can activate the signal and propagate the signal to the downstream neurons connected to it, which is usually done based on whether the combined received signal, which may be from many transmitting neurons, is of sufficient strength, and the strength is a parameter.

[0024] A deep neural network (DNN) is a stacked neural network composed of multiple layers. Each layer consists of nodes, which are where computations take place and are loosely patterned on neurons in the human brain, activating when they encounter sufficient stimulation. Nodes combine inputs from data with a set of coefficients and / or weights to amplify or attenuate that input, assigning importance to the input for the task the algorithm is trying to learn. The products of these inputs and weights are summed and passed to the node's so-called activation function to determine whether, and to what extent, the signal travels further through the network and influences the final result. DNNs use a cascade of many layers of nonlinear processing units for feature extraction and transformation. Each subsequent layer uses the output from the previous layer as input. Higher-level features are derived from lower-level features to form a hierarchical representation. The layer following the input layer may be a convolutional layer that generates a feature map, which is the result of filtering the input and is used by the next convolutional layer.

[0025] The DNN may be a specific type of DNN, such as a Convolutional Neural Network (CNN), a Recurrent Neural Network (RNN), or a Long-Short-Term Memory (LSTM). In some examples, other artificial neural networks may be used. In some examples, a classifier may be used instead of a neural network. A classifier does not need to include hidden layers, but it can classify a particular input as corresponding to a particular output. For example, for a set of pressure and / or flow rate data, a classifier may produce a distinction between CV and non-CV.

[0026] Input data for training Model 202 may include data captured from a CVQS (e.g., CVQS100 in Figure 1), along with labeled data from physicians and / or patients. Model 202 may be used in an inference stage (more detailed below with respect to Figure 3) to determine the presence (or absence) of collateral ventilation and / or to indicate whether the patient is a candidate for a lumen valve.

[0027] As shown in Figure 2, training data may include signal training data consisting of measurement signals representing quantifiable measurements obtained by the CVQS's flowmeter and / or pressure sensors. For example, training data may include measurements of airflow to and from a lobe that is occluded by the CVQS and is being considered for treatment with the placement of a lumen valve in one or more airways that are in fluid communication with the occluded lobe. It will be understood that, because the lobe is occluded, the airflow entering and leaving the occluded lobe through the occluded air passage will pass entirely through the lumen of the CVQS, bypassing the occluding device (e.g., assuming the occluding device forms a perfect seal, which in reality does not always occur in real-life environments). Signal training data may include data provided by a CPAP machine actively ventilating the patient (e.g., constant-pressure air supply device 112 in Figure 1). In some examples, training data may include annotation training data provided by the physician and / or patient (e.g., as labeling data). For example, a physician may annotate a patient-specific CVQS dataset as indicating the presence or absence of CV, and / or as representing the degree of assessment of CV and / or collateral resistance (Rcoll). This annotated training data may be used to train Model 202. Additionally or alternatively, patient-specific CVQS datasets may be annotated with subjective patient outcome feedback after valve implantation. For example, individual patients may be assessed through CVQS behavior to generate a patient-specific CVQS dataset. Following this assessment, one or more luminal valves may be implanted in the patient's body, and following this valve implantation procedure (e.g., one week after valve implantation, one month after valve implantation, etc.), the patient may provide subjective feedback on perceived improvement or lack thereof.Patient outcome feedback may include a sliding scale value of perceived life improvement resulting from valve implantation (for example, how previously reported dyspnea improved after the valve implantation procedure, on a scale of 1 to 10).

[0028] Based on signal training data and / or annotation training data, Model 202 can generate output weights corresponding to individual processing nodes spread across the input layer, output layer, and one or more hidden layers. Model 202 and the trained weights can later be used to infer, based on new inputs from the patient under consideration, whether the patient is CV+ or CV-, whether the degree of CV is indicated, and / or whether the patient is suitable for a predetermined type of treatment (e.g., luminal valve implantation).

[0029] Figure 3 shows a machine learning model inference diagram 300, according to at least one example of the present disclosure. In the inference diagram 300, model 302 (e.g., model 202 after training and / or update) may be used to output predictions such as whether a patient has a central venous cavity (CV) and whether treatment is recommended (e.g., a luminal valve). Confidence levels and / or weights may be output as predictions or in addition to other predictions discussed above. The machine learning model inference diagram 300 may represent an exemplary computer-based clinical decision support system (CDSS) configured to assist in predicting patient-specific outcomes resulting from the implantation of one or more luminal valves into one or more bronchial airways that fluidly communicate with the affected portion of the lungs of a COPD patient.

[0030] As shown in Figure 3, Model 302 may receive signals as input from a CVQS (e.g., CVQS100 in Figure 1), such as pressure data, flow rate data, and / or data provided by a CPAP machine (e.g., constant pressure air supply device 112 in Figure 1), and / or other data such as patient data. Model 302 may generate outputs (e.g., inferences) that include a patient suitability assessment (e.g., CV present or CV absent), predicted patient outcomes (e.g., an indication of what might happen if valve implantation is performed based on the currently observed input signals), confidence levels (e.g., 95% confidence in the patient suitability assessment "CV absent", 95% confidence in valve implantation reducing dyspnea, etc.), and estimated CV volume. Model 302 may have execution times that occur while the patient is undergoing and / or immediately after completion of the CVQS procedure. Model 302 can provide physicians with a rapid on-site assessment of whether the current patient is undergoing collateral ventilation and / or whether the current patient could benefit from valve placement in an obstructed lobe.

[0031] Figure 4A shows a flow and pressure diagram illustrating an example of collateral ventilation, according to at least one example of the present disclosure. Figure 4A includes an arrow indicating whether balloon occlusion occurred (e.g., at 20 seconds). The indicator of CV is not readily visible to an untrained eye, and in some examples, may not be detectable even by a trained physician. In one example, the output graph shown in Figure 4A may be used as input to a machine learning-trained model to determine whether the corresponding patient has CV. The model may use a classifier and / or a neural network to determine from the graph and / or underlying data whether CV is present.

[0032] Figure 4B shows a flow and volume diagram illustrating an example of collateral ventilation according to at least one example of the present disclosure. The vertical graph portion represents flow rate data, and the increasing line represents total flow rate.

[0033] Figure 4C shows a flow and pressure diagram illustrating an example of collateral ventilation absence (no CV) according to at least one example of this disclosure. Figure 4D shows a flow and volume diagram illustrating an example of collateral ventilation absence (no CV) according to at least one example of this disclosure. In the absence of collateral ventilation, the airflow weakens and / or stops when the pressure in the obstructed lobe reaches the original pressure. If the ventilator is paused during the performance of the assessment, air continues to flow into the lobe until it is pressurized to the CPAP pressure. The graphs shown in Figures 4C–4D represent the (CV negative, no CV) condition.

[0034] Figure 5 shows a flowchart illustrating technique 500 for training a model to determine whether a patient is a candidate for lumen valve implantation based on collateral ventilation data, according to at least one example of the present disclosure. Technique 500 may be executed by a processor by executing instructions stored in memory.

[0035] Technique 500 includes operation 502, which receives data, for example, captured by a sensor, based on pressure and / or airflow in a target portion of the lung of a patient whose respiratory tract is obstructed so that it cannot receive air through the respiratory tract of the lung. The obstructed respiratory tract may be obstructed by a balloon, which may block the outflow tract. In this example, the received data may be pressure data based on positive pressure applied to the inflow tract, for example, a constant pressure of 10 cmH2O. In some examples, the obstructed respiratory tract is obstructed by a valve that blocks the inflow tract while allowing outflow air, and the received data is outflow air data. In one example, operation 502 includes periodically acquiring measured data of airflow and / or pressure in a target portion of the lung.

[0036] Technique 500 includes an operation 504 to label received data based on corresponding patient respiratory outcomes in order to generate training data and / or receive labeled data. In one example, the corresponding patient respiratory outcome includes a clinician's determination, based on received data, whether the patient is performing collateral ventilation in a target portion of the lung. In another example, the corresponding patient respiratory outcome includes objective metrics of the patient's respiration and / or a patient-reported respiratory assessment taken after a procedure to insert a lumen valve into the patient's body. In some examples, a combination of corresponding patient respiratory outcomes may be used.

[0037] Technique 500 includes operation 506 for training a machine learning model based at least in part on training data to predict a patient's respiratory outcome through an indication of whether collateral ventilation is present in a target lung portion of a particular patient. Operation 506 may include a step of using at least one of the following as additional input data: volume data, a medical image of the patient, a crack integrity score, the patient's disease status, the patient's age, and / or the patient's comorbidities. The indication may include a binary indication of either collateral ventilation positive or collateral ventilation negative (for example, via a user interface displaying text and / or images, such as green light for CV positive or red light for CV negative). The indication may include the likelihood, confidence level, etc., that the patient is collateral ventilation positive.

[0038] Technique 500 includes an action 508 for outputting a machine learning model. Action 508 may include deploying the machine learning model (e.g., making the machine learning model available via API, the internet, download, etc.), saving the machine learning model (e.g., for retrieval and / or updating for later use), and sending the machine learning model to a destination (e.g., a database and / or server).

[0039] Technique 500 may include actions to obstruct the respiratory airway in a target portion of the lung (e.g., using a valve, balloon, etc.).

[0040] Figure 6 is a block diagram of an exemplary machine 600 in which one or more of the techniques discussed herein may be performed according to several embodiments. In alternative embodiments, machine 600 may operate as a standalone device and / or be connected to other machines (e.g., network connection). In a networked deployment, machine 600 may operate as a server machine, a client machine, or both in a server-client network environment. In one example, machine 600 may function as a peer machine in a peer-to-peer (P2P) (or other distributed) network environment. Machine 600 may be a personal computer (PC), tablet PC, set-top box (STB), personal digital assistant (PDA), mobile phone, web appliance, network router, switch or bridge, or a machine capable of executing instructions (sequential or otherwise) specifying the actions it should perform. Furthermore, although only a single machine is shown, the term “machine” should be interpreted to include any set of machines individually or collectively performing one or more of the methodologies described herein, such as cloud computing, software as a service (SaaS), and other computer cluster configurations.

[0041] The machine (e.g., a computer system) 600 may include a hardware processor 602 (e.g., a central processing unit (CPU), a graphics processing unit (GPU), a hardware processor core, or any combination thereof), main memory 604, and static memory 606, some or all of which may communicate with each other via an interlink (e.g., a bus) 608. The machine 600 may further include a display unit 610, an alphanumeric input device 612 (e.g., a keyboard), and a user interface (UI) navigation device 614 (e.g., a mouse). In one example, the display unit 610, the input device 612, and the UI navigation device 614 may be touchscreen displays. The machine 600 may further include a storage device (e.g., a drive unit) 616, a signal generating device 618 (e.g., a speaker), a network interface device 620, and one or more sensors 621 such as a Global Positioning System (GPS) sensor, a compass, an accelerometer, or other sensors. The machine 600 may include an output controller 628, such as a serial (e.g., Universal Serial Bus (USB)), parallel, or other wired or wireless (e.g., infrared (IR), near-field communication (NFC)) connection for communicating with and / or controlling one or more peripheral devices (e.g., a printer, a card reader, etc.).

[0042] The storage device 616 may include a machine-readable medium 622 in which one or more sets of data structures or instructions 624 (e.g., software) that embody or utilize any one or more of the techniques or functions described herein are stored. The instructions 624 may also reside, all or at least partially, in main memory 604, static memory 606, or hardware processor 602 during execution by machine 600. In one example, one or any combination of hardware processor 602, main memory 604, static memory 606, or storage device 616 may constitute the machine-readable medium.

[0043] Although machine-readable medium 622 is shown as a single medium, the term “machine-readable medium” may include a single or multiple mediums configured to store one or more instructions 624 (for example, a centralized or distributed database, and / or associated caches and servers). The term “machine-readable medium” may include any medium capable of storing, encoding, or carrying instructions for execution by machine 600, causing machine 600 to execute one or more of the techniques of the Disclosure, or capable of storing, encoding, or carrying data structures used by or associated with such instructions. Non-limiting examples of machine-readable mediums may include solid-state memory, optical media, and magnetic media.

[0044] Instruction 624 may be further transmitted or received over the communication network 626 using a communication medium via the network interface device 620, utilizing one of a number of transport protocols (e.g., Frame Relay, Internet Protocol (IP), Transmission Control Protocol (TCP), User Datagram Protocol (UDP), Hypertext Transfer Protocol (HTTP), etc.). Illustrative communication networks may include, among others, local area networks (LANs), wide area networks (WANs), packet data networks (e.g., the Internet), mobile telephone networks (e.g., cellular networks), plain old telephone (POTS) networks, and wireless data networks (e.g., the IEEE 802.11 standard family known as Wi-Fi®, the IEEE 802.16 standard family known as WiMax®), the IEEE 802.15.4 standard family, peer-to-peer (P2P) networks, etc. For example, the network interface device 620 may include one or more physical jacks (e.g., Ethernet, coaxial, or telephone jacks) or one or more antennas for connecting to the communication network 626. For example, the network interface device 620 may include multiple antennas for wireless communication using at least one of the single-input multiple-output (SIMO), multiple-input multiple-output (MIMO), or multiple-input single-output (MISO) techniques. The term “transmitting medium” is interpreted to include any intangible medium on which instructions for execution by the machine 600 can be stored, encoded, or carried, including digital or analog communication signals or other intangible mediums that facilitate the communication of such software.

[0045] Each of the following non-restrictive examples may exist on its own or may be combined with one or more of the other examples in various permutations or combinations.

[0046] Example 1 is a collateral ventilation quantification system for training a machine learning model for use in a computer-based clinical decision support system to help predict outcomes for patients with luminal valve implantation, the collateral ventilation quantification system comprising: at least one sensor for capturing data based on at least one of pressure or airflow in a target portion of a patient's lung that is obstructed by a device so that it cannot receive air through the respiratory airway of the lung; a processing circuit; and a memory containing instructions that, when performed by the processing circuit, cause the processing circuit to perform operations including labeling the received data based on the corresponding patient's respiratory outcome to generate training data; training a machine learning model at least partially on the training data to predict the respiratory outcome of one or more patients via an indicator of whether collateral ventilation is present in a particular patient's target lung portion; and storing the machine learning model.

[0047] Example 2 is a method for training a machine learning model for use in a computer-based clinical decision support system to help predict outcomes for patients with luminal valve implantation, the method comprising: receiving data captured by at least one sensor indicating at least one of pressure or airflow in a target portion of a patient's lung that is obstructed by a device so that it cannot receive air through the respiratory airway of the lung; labeling the received data based on the corresponding patient's respiratory outcome to generate training data; training a machine learning model at least in part on the training data to predict the respiratory outcome of one or more patients via an indicator of whether collateral ventilation is present in a particular patient's target lung portion; and outputting the machine learning model.

[0048] In Example 3, the subject of Example 2 is that the obstructed respiratory airway is blocked by a balloon, thereby blocking the outflow airway, and the received data is pressure data based on positive pressure applied to the inflow airway.

[0049] In Example 4, the subject of Example 3 includes the fact that the applied positive pressure includes a constant applied pressure.

[0050] In Example 5, the subject matter of Examples 2-4 includes training a machine learning model using at least one of the following as additional input data: lung volume data, patient medical images, crack integrity scores, patient disease status, patient age, or patient comorbidities.

[0051] In Example 6, the subject matter of Examples 2-5 includes the fact that the corresponding patient's respiratory outcome includes the clinician's determination, based on received data, whether the patient is performing collateral ventilation in the target portion of the lung.

[0052] In Example 7, the subject matter of Examples 2-6 includes the corresponding patient respiratory outcome, which includes objective results of the patient's respiration or a respiratory assessment reported by the patient after a procedure to insert a lumen valve into the patient's body.

[0053] In Example 8, the subject matter of Examples 2-7 includes the fact that an indicator of whether collateral ventilation is present in a target lung portion of a particular patient is output from the model as a binary representation of whether collateral ventilation is present or not.

[0054] In Example 9, the subject matter of Examples 2-8 includes the output of a mark from the model that includes the probability that the patient is performing collateral ventilation in the target area.

[0055] In Example 10, the subject matter of Examples 2–9 involves using a device to obstruct the respiratory airway of a target portion of the lung.

[0056] In Example 11, the subject of Examples 2-10 includes a step of receiving data, which involves repeatedly or periodically acquiring measurement data of airflow or pressure in a target portion of the lung.

[0057] In Example 12, the subject matter of Examples 2-11 includes the fact that an obstructed respiratory airway is blocked by a valve that blocks the inflow airway while allowing outflow air, and the received data is outflow air data.

[0058] Example 13 is a method for training a machine learning model for use in a computer-based clinical decision support system to help predict outcomes for patients with luminal valve implantation, the method comprising: receiving pressure data captured by at least one sensor indicating pressure in a target portion of the lung of a patient that is obstructed by a device so that it cannot receive air through the respiratory airway of the lung; labeling the received pressure data based on the corresponding patient's respiratory outcome to generate training data; training a machine learning model at least in part on the training data to predict the respiratory outcome of one or more patients via an indicator of whether collateral ventilation is present in a particular patient's target lung portion; and storing the machine learning model.

[0059] Example 14 is a method for training a machine learning model for use in a computer-based clinical decision support system to help predict outcomes for patients with luminal valve implantation, the method comprising: receiving airflow data captured by at least one sensor indicating airflow from a target portion of a patient's lung that is obstructed by a device so that it cannot receive air through the respiratory airway of the lung; labeling the received airflow data based on the corresponding patient's respiratory outcome to generate training data; training a machine learning model at least in part on the training data to predict respiratory outcomes for one or more patients via an indicator of whether collateral ventilation is present in a particular patient's target lung portion; and storing the machine learning model.

[0060] Example 15 is a device for training a machine learning model for use in a computer-based clinical decision support system to help predict outcomes for patients undergoing luminal valve implantation, the device comprising: a processing circuit; a memory including instructions that, when executed by the processing circuit, cause the processing circuit to receive data captured by at least one sensor indicating pressure or airflow in a target portion of a patient's lung that is obstructed by the device so that it cannot receive air through the respiratory airway of the lung; and instructions that cause the processing circuit to perform operations including labeling the received data based on the corresponding patient's respiratory outcome to generate training data; training a machine learning model at least partially on the training data to predict the respiratory outcome of one or more patients via an indicator of whether collateral ventilation is present in a particular patient's target lung portion; and outputting the machine learning model.

[0061] Example 16 is at least one machine-readable medium which, when executed by the processing circuit, causes the processing circuit to perform an operation that includes receiving data captured by at least one sensor indicating pressure or airflow in a target portion of a patient's lung that is obstructed by a device so that it cannot receive air through the respiratory airway of the lung; labeling the received data based on the corresponding patient's respiratory outcome to generate training data; training a machine learning model at least partially on the training data to predict the respiratory outcome of one or more patients via an indicator of whether collateral ventilation is present in a particular patient's target lung portion; and outputting the machine learning model.

[0062] Example 17 is a method comprising the steps of: receiving data captured by at least one sensor indicating at least one of pressure or airflow in a target portion of a patient's lung that is obstructed by a device so that it cannot receive air through the respiratory airway of the lung; implementing a machine learning model trained at least in part on training data including input previous patient sensor data and corresponding previous patient respiratory results labeled, in order to predict the patient's respiratory outcome; and outputting at least one indicator based on the prediction from the machine learning model, which is either that collateral ventilation is present or that the predicted patient respiratory outcome corresponds to the placement of a lumen valve in the patient's body.

[0063] In Example 18, the subject of Example 17 includes the step of outputting a mark, which includes the step of identifying the presence of collateral ventilation and, accordingly, displaying a recommendation to treat the patient with a lumen valve.

[0064] In Example 19, the subject of Examples 17–18 includes a step of outputting a mark, which includes identifying the absence of collateral ventilation and, accordingly, displaying a recommendation not to treat the patient with a lumen valve.

[0065] In Example 20, the subject matter of Examples 17–19 includes the fact that an obstructed respiratory airway is occluded by a balloon to block the outflow airway, and the received data is pressure data based on positive pressure applied to the inflow airway.

[0066] In Example 21, the subject of Example 20 includes the fact that the applied positive pressure includes a constant applied pressure.

[0067] In Example 22, the subject of Examples 17-21 includes a step of outputting a mark, which in turn outputs the probability that the patient is performing collateral ventilation in the target area.

[0068] In Example 23, the subject matter of Examples 17–22 involves using a device to obstruct the respiratory airway of a target portion of the lung.

[0069] In Example 24, the subject of Examples 17–23 includes a step of receiving data, which involves repeatedly or periodically acquiring measurement data of airflow or pressure in a target portion of the lung.

[0070] In Example 25, the subject matter of Examples 17-24 includes the fact that an obstructed respiratory airway is blocked by a valve that blocks the inflow airway while allowing outflow air, and the received data is outflow air data.

[0071] Example 26 is at least one machine-readable medium that, when executed by a processing circuit, contains instructions causing the processing circuit to perform an action to implement any of Examples 1 to 25.

[0072] Example 27 is a device that includes means for implementing any of Examples 1 to 25.

[0073] Example 28 is a system for implementing any of Examples 1 through 25.

[0074] Example 29 is a method for implementing any of Examples 1 through 25.

[0075] Examples of methods described herein may be implemented, at least in part, by machine or by computer. Some examples may include computer-readable or machine-readable media encoded with instructions that can be used to configure an electronic device to perform the methods described above. Implementations of such methods may include code such as microcode, assembly language code, or high-level language code. Such code may include computer-readable instructions for performing various methods. The code may form part of a computer program product. Furthermore, in one example, the code may be tangibly stored in one or more volatile, non-temporary, or non-volatile tangible computer-readable media during execution or at any other time. Examples of these tangible computer-readable media may include, but are not limited to, hard disks, removable magnetic disks, removable optical disks (e.g., compact disks and digital video disks), magnetic cassettes, memory cards or memory sticks, random access memory (RAM), read-only memory (ROM), and the like. [Explanation of symbols]

[0076] 100 Collateral Ventilation Quantification System (CVQS) 102 CVQS devices 104 CVQS Tubing Kit 106 Flow meter 108 Pressure Gauge 110 Display Devices 112 Constant pressure air supply device 114 Check valve 116 Obstruction devices Training diagrams of 200 machine learning models 202 Model 300 Machine Learning Model Inference Diagrams 302 Model 500 techniques 502 operation 600 machines 602 Hardware Processors 604 Main Memory 606 Static Memory 608 Interlink 610 Display Unit 612 alphanumeric input devices 614 User Interface (UI) Navigation Devices 616 storage devices 618 Signal Generating Devices 620 Network Interface Devices 621 Sensor 622 Machine-readable media 624 Command 626 Communication Network 628 Output Controller

Claims

1. A collateral ventilation quantification system for training machine learning models for use in a computer-based clinical decision support system to help predict outcomes in patients undergoing luminal valve implantation, wherein the collateral ventilation quantification system is At least one sensor for capturing data based on at least one of pressure or airflow in a target portion of the lung of a patient whose lung is obstructed by the device so that it cannot receive air through the respiratory ducts of the lung, Processing circuit and When executed by the processing circuit, the processing circuit will Receiving the data from the aforementioned sensor, The aforementioned data is labeled with an indicator of whether or not collateral ventilation was present, and training data is generated. To train a machine learning model based at least partially on the training data in order to predict the respiratory outcome of one or more patients resulting from lumen valve intervention, based on whether or not collateral ventilation was present, The aforementioned machine learning model is stored and Memory containing instructions that cause an operation including A system that includes these features.

2. A method for training a machine learning model for use in a computer-based clinical decision support system to help predict outcomes in patients undergoing luminal valve implantation, The steps include receiving data captured by at least one sensor indicating at least one of the pressure or airflow in a target portion of the lung that is obstructed by the device so that it cannot receive air through the respiratory airway of the patient's lung, The steps include labeling the aforementioned data with an indicator of whether or not collateral ventilation was present, and generating training data, The steps include training a machine learning model on the training data, at least partially, to predict the respiratory outcome of one or more patients resulting from lumen valve intervention, based on whether or not collateral ventilation was present, The step of outputting the aforementioned machine learning model Methods that computers perform, including those mentioned above.

3. The method according to claim 2, wherein the obstructed respiratory airway is blocked by a balloon to block the outflow airway, and the received data is pressure data based on positive pressure applied to the inflow airway.

4. The method according to claim 3, wherein the applied positive pressure includes a constant applied pressure.

5. The method according to claim 2, wherein the step of training the machine learning model includes using at least one of the following as additional input data: lung volume data, medical images of the patient, crack integrity score, the patient's disease status, the patient's age, or the patient's comorbidities.

6. The method according to claim 2, wherein the respiratory results of the corresponding patient include a clinician's determination, based on the received data, whether the patient is performing collateral ventilation in the target portion of the lung.

7. The method according to claim 2, wherein the corresponding patient's respiratory results include objective results of the patient's respiration, or a respiratory assessment reported by the patient obtained after a procedure to insert a lumen valve into the patient's body.

8. The method according to claim 2, wherein the indicator of whether collateral ventilation is present in a target portion of the lung of a particular patient is output from the machine learning model as a binary representation of whether collateral ventilation is present or not.

9. The method according to claim 2, wherein the machine learning model outputs a mark that includes the probability that the patient is performing collateral ventilation in the target area.

10. The method according to claim 2, further comprising the step of using the device to occlude the respiratory airway of the target portion of the lung.

11. The method according to any one of claims 2 to 10, wherein the step of receiving the data includes the step of repeatedly or periodically acquiring measurement data of the airflow or pressure in the target portion of the lung.

12. The method according to any one of claims 5 to 10, wherein the obstructed respiratory airway is blocked by a valve that blocks the inflow airway while allowing outflow air, and the received data is outflow air data.

13. The steps include receiving data captured by at least one sensor indicating at least one of the pressure or airflow in a target portion of the lung that is obstructed by the device so that it cannot receive air through the respiratory airway of the patient's lung, To predict the respiratory outcome of the patient, the steps include: implementing a machine learning model trained at least partially on training data that includes at least one of (i) the corresponding patient respiratory outcome associated with luminal valve intervention, or (ii) an indicator of whether or not collateral ventilation was present; The steps include inputting the data captured by at least one sensor into the machine learning model, (i) outputting an indicator of whether collateral ventilation is present in a target portion of a particular patient's lung, based on at least one of the corresponding patient's respiratory outcome associated with the luminal valve intervention, or (ii) an indicator of whether collateral ventilation was present. A step of predicting the respiratory outcome of one or more patients for the aforementioned patient. Methods that computers perform, including those mentioned above.

14. The method according to claim 13, wherein the step of outputting the indicator includes the step of identifying the presence of collateral ventilation and, accordingly, displaying a recommendation to treat the patient with a luminal valve.

15. The method according to claim 13, wherein the step of outputting the aforementioned indicator includes the step of identifying the absence of collateral ventilation and, accordingly, displaying a recommendation not to treat the patient with a lumen valve.

16. The method according to claim 13, wherein the obstructed respiratory airway is blocked by a balloon to block the outflow airway, and the received data is pressure data based on positive pressure applied to the inflow airway.

17. The method according to claim 16, wherein the applied positive pressure includes a constant applied pressure.

18. The method according to claim 13, wherein the step of outputting the mark includes the step of outputting the probability that the patient is performing collateral ventilation in the target area.

19. The method according to claim 13, further comprising the step of using the device to occlude the respiratory airway of the target portion of the lung.

20. The method according to any one of claims 13 to 19, wherein the step of receiving the data includes the step of repeatedly or periodically acquiring measurement data of the airflow or pressure in the target portion of the lung.

Images